<div dir="ltr">Hi,<div> </div><div>While I too have questions about the annealing procedure in CP2K, if you have a small system, Nose-Hoover thermostatics is a generally bad idea regardless of its time constant. </div><div>Try CSVR or adaptive Langevin. For the initial relaxation, try to go with a very tight temperature control (a time constant of a few femtoseconds for CSVR, for example).</div><div> </div><div>Best,</div><div> </div><div>Alex On Wednesday, November 5, 2014 2:50:30 AM UTC-7, Bouzid Assil wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div dir="ltr">Dear CP2k users, I am using cp2k to perform MD simulations based on the second generation Car-Parrinello MD (Kühne scheme). I have started by a small binary chalcogenide system that I have already generated using the standard CPMD. As a first step I want to relax my system at 0K using the ANNEALING procedure, this is supposed to reduce the forces acting between the "atoms" at each MD step by a given factor (say 0.9). To control the situation I have used at first gamma_D = 0.0000. My problem is that after more than 30ps the temperature stabilize around 1000K and won't go down. I have also tried an NVT run at 300K using the NOSE HOOVER thermostat but no way, the thermostat is not able to control the temperature (I have tried different TIMECON values, always the same behaviour!). When analysing the output files, there is a common strange thing between all these runs (and also an other run on water molecules), when performing the SCF WAVEFUNCTION OPTIMIZATION typically the convergence is reached after 2 iterations of the outer SCF loop BUT after the first ~10 MD steps the convergence is not reached at all and only 2 OT steps are performed! is this normal (for sure no!) but why it happens then?! here is an example of output: ----------------------------------- OT --------------------------------------- Allowing for rotations: F Optimizing orbital energies: F Minimizer : DIIS : direct inversion in the iterative subspace using : - 7 DIIS vectors - safer DIIS on Preconditioner : FULL_KINETIC : cholesky inversion of T + eS Precond_solver : DEFAULT stepsize : 0.08750000 energy_gap : 0.00200000 eps_taylor : 0.10000E-15 max_taylor : 4 mixed_precision : F ----------------------------------- OT --------------------------------------- Step Update method Time Convergence Total energy Change ------------------------------------------------------------------------------ 1 OT DIIS 0.87E-01 2.1 0.00000982 -906.5667899239 -9.07E+02 2 OT DIIS 0.87E-01 2.5 0.00000932 -906.5667939977 -4.07E-06 *** SCF run NOT converged *** Electronic density on regular grids: -640.0000000000 -0.0000000000 Core density on regular grids: 639.9999999985 -0.0000000015 Total charge density on r-space grids: -0.0000000015 Total charge density g-space grids: -0.0000000015 Overlap energy of the core charge distribution: 0.00000056633200 Self energy of the core charge distribution: -1927.34045098667070 Core Hamiltonian energy: 434.65748894621680 Hartree energy: 790.75025229030484 Exchange-correlation energy: -204.63408481387373 Total energy: -906.56679399769098 ENERGY| Total FORCE_EVAL ( QS ) energy (a.u.): -906.566801465010712 ******************************************************************************* ENSEMBLE TYPE = LANGEVIN STEP NUMBER = 34788 TIME [fs] = 34788.000000 CONSERVED QUANTITY [hartree] = -0.906018037069E+03 INSTANTANEOUS AVERAGES CPU TIME [s] = 9.85 9.56 ENERGY DRIFT PER ATOM [K] = -0.352815792553E+06 -0.346958228435E+06 POTENTIAL ENERGY[hartree] = -0.906566801465E+03 -0.906008788692E+03 KINETIC ENERGY [hartree] = 0.548764395757E+00 0.221663346691E+01 TEMPERATURE [K] = 970.789 3921.326 ******************************************************************************* Here is a part of my input file: &MOTION &MD ENSEMBLE LANGEVIN STEPS 15000 ANNEALING 0.8 TIMESTEP 1.0 &LANGEVIN GAMMA 0.0001 NOISY_GAMMA 0.000 &END LANGEVIN &END MD &PRINT &RESTART &EACH MD 1 &END EACH BACKUP_COPIES 3 &END RESTART &END PRINT &END MOTION &FORCE_EVAL METHOD QUICKSTEP &DFT BASIS_SET_FILE_NAME BASIS_MOLOPT POTENTIAL_FILE_NAME GTH_POTENTIALS WFN_RESTART_FILE_NAME meta-RESTART.wfn &MGRID &RS_GRID DISTRIBUTION_TYPE REPLICATED &END RS_GRID CUTOFF 300 &END MGRID &PRINT &DFT_CONTROL_PARAMETERS &END DFT_CONTROL_PARAMETERS &END PRINT &QS EPS_DEFAULT 1.0e-12 EXTRAPOLATION ASPC EXTRAPOLATION_ORDER 3 MAP_CONSISTENT TRUE &END QS &SCF EPS_SCF 1.0e-7 MAX_SCF 50 MAX_SCF_HIST 2 &OT ON MINIMIZER DIIS PRECONDITIONER FULL_KINETIC ENERGY_GAP 0.002 STEPSIZE 0.0875 &END OT &OUTER_SCF ON MAX_SCF 100 EPS_SCF 1.0e-7 &END OUTER_SCF Any input on the possible origin of this problem is highly appreciated! Many thanks. All the best, Assil Bouzid </div></blockquote></div></div>